Hi. Thanks for posting this here instead of an email. You've asked some good questions I think others can benefit from!
If my understanding is correct, the VGS(th) from the datasheet gives you the gate voltage where the transistor will open and become active.
Vgs(th) is the minimum voltage for the MOSFET to turn on (or off really). That threshold voltage generally represents the worst-case (or most) resistance for RdsON. So if the data sheet says "35mOhm RdsON" in the features section, it is going to be much higher at the minimum threshold voltage.
The range is just that, the specified range. MOSFETs really usually meant to be driven at their threshold voltage. So a transistor with a range of 2 - 5 V probably should be driven with more than 5 V.
How do I need to add the Rds(on) resistance in my calculation.
Treat it like a series resistor.
But as you mentioned at some point the MOSFET becomes an adjustable resistor. Wouldn't that mean that the resistance should be able to rise way higher?
Enhancement Mode MOSFETs, like the FQP, do not vary much. The RdsON curve in the datasheet shows it. It will be something like 30mOhm to 100mOhm. Not really enough to call it a true variable resistor. With an application like driving a LED, you won't notice the brightness difference between within that range.
The reason I used the phrase "variable resistor" in the video was to help show the difference from BJTs. (JFETs, on the other hand, are much more variable.)
How do I have to understand the Figure 1 "On-Region Characteristics" diagram from the datasheet?
That graph condenses some parameters into a single chart. You have Vgs voltages along with drain current and voltage drops from drain to source. In other words, they've calculated what RdsON will be and are showing potential power dissipation, based on the drain current.
How could I get a Vds of 1,3V with a Vgs of 3.5V and like 2A if the Vds voltage should drop to nearly 0V when conducting? (like the Vce of a BJT)
Because RdsON is the resistance from Drain to Source. So there is some resistance, which means there will be a voltage drop. I don't follow how you got 1.3 V. The graph for a FQP30N06L shows about Vds at 0.15 V with Vgs at 3.5 V and Id at 2 A. Which Ohm's law would mean RdsON at that point is 75mOhm.
replace the BJT with the MOSFET and put a gate resistor of ~1MOhm between gate and ground
I would recommend a resistor from gate to ground, to "pull-down" the gate. But I'm not sure if that is what you're asking. You should include a resistor between the gate and whatever is turning the gate on or off. That is meant to limit the current into the MOSFET since the gate is basically a very small capacitor.
Lets say my LED draws 20mA
The "current rating" for an LED is the maximum rating. It is also the value which will give the LED its shortest (rated) life. Just because the LED can handle 20 mA, does not mean you have to drive it that high. Lower current means longer operational life.
Could you tell me where or for what applications you typical use MOSFETs, BJTs or even JFETs?
As I said in the video, in general (and many, many people ignored that one word), MOSFETs are used in high current, high switching frequency applications. BJTs are typically used in small single applications. For high-voltage applications, like the inverter for an electric vehicle's motor, you'll see an IGBT. Which, in simple terms, is a combination of BJT and MOSFET.
In applications where it matters which you use, the characteristics of the two kinds of transistors makes it obvious which to use. In cases where it doesn't, engineers will give you opinions as facts to which you should be using.
Like should I use a switching regulator a linear regulator or a voltage divider with an impedance follower opamp to power my microcontroller?
Like all things in electronics, it depends. I don't think you should ever use a voltage divider and opamp to power a microcontroller. If you're not doing any analog to digital with a built in ADC, then switching is ideal for power and efficiency. Linear is ideal for the cost. Engineering is evaluating trade-offs for a particular design. "Rules" for one design may not apply to another. Sometimes, you just have to learn from experience.